Another Senate Prediction Glows Red

An excellent prediction model from 2012 reappears for the midterm elections.
He's not surrendering.

Welcome to yet another Senate prediction model (state charts here) -- this one from Daily Kos. Basically, this is political scientist Drew Linzer’s “Votamatic” system from the 2012 presidential election, now applied to the 2014 Senate races. Linzer called every state correctly in 2012; more important, since a lot of luck goes into calls in very close races, he has excellent statistical chops and his methodology is open to outside inspection. So it’s good stuff.

Unlike the Monkey Cage/Washington Post system, the Upshot system at the New York Times or 538, Linzer's system doesn’t evaluate the “fundamentals” -- presidential approval ratings, economic growth, etc. It simply assesses the relevant polls. (For other models, see Larry Sabato’s Crystal Ball, or the Cook and Rothenberg reports.)

It’s helpful to have so many windows on the election cycle, and right now they’re all telling us the same thing: Control of the next Senate is a tossup with a slight or better-than-slight tilt toward the Republicans. Specifically, Daily Kos shows a 53 percent chance that Republicans will hold 51 or more seats in the next Senate, while the Post gives Republicans a 61 percent chance and the Times predicts a 64 percent likelihood.

Linzer's (Daily Kos) system doesn't come in at the low end because of a Democratic bias. It’s because the polls look better for Democrats than the fundamentals. Indeed, the Post prediction was more bullish on a Republican Senate takeover before it started incorporating polling. Right now, in other words, Republicans are running short of expectations given the public's unenthusiastic views of a second-term Democratic president and the economy, the particular seats in play during this cycle, and in light of fairly successful candidate recruitment efforts by Republicans.

The most likely result, according to the Daily Kos method, would yield either 50 or 51 Republican seats. But anything from 47 to 54 Republican seats would still fit the model. That’s basically true for the other systems as well. No one knows which way the closest races will go. And I’ll say it again: as important as majority control is, in the Senate each individual seat matters quite a bit.

The outcomes of those close races could be influenced by local context and events, including campaign gaffes, great advertising or the differences in quality between get-out-the-vote operations. Or they could be determined by small shifts at the national level, involving war and peace or the economy, for example. Or it could be some combination of circumstances.

At any rate, Linzer’s Votamatic was a great tool in 2012; I’m glad to see it revived for the 2014 cycle.

This column does not necessarily reflect the opinion of Bloomberg View's editorial board or Bloomberg LP, its owners and investors.